This is a renewal R01 application to exploit knock-in mouse models and hypermutating human B cell lines carrying germline-reverted broadly neutralizing antibodies (gl-bNAbs) against HIV Env with the goal to speed HIV vaccine research. Developing an HIV vaccine is a major global health objective as no effective vaccine yet exists. bnAbs to HIV Env glycoprotein can likely control infection but eliciting bnAbs by vaccination is challenging because of Env genetic variability and the fact that responses to conserved Env epitopes are weak. bnAbs are made naturally in some patients, but usually only after years of infection, when they provide no real protection. However, many individual bnAbs can provide passive protection in animal models, suggesting that elicitation of these antibodies could be protective, particularly if bnAbs to several conserved sites can be raised simultaneously. To begin to investigate the best ways to elicit bnAbs by vaccination, we have previously generated B cell lines and knock-in mice carrying inferred gl-bnAbs for VRC01, PGT121, b12, 4E10 and other specificities. gl-bnAb mouse models are valuable for antigen design and vaccination studies because they carry B cells with the potential to become bnAbs after appropriate stimulation and somatic mutation. Placement of the V(D)J genes into the physiological Ig loci allows the knock-in B cells to undergo normal H-chain class switching and V-region hypermutation. Moreover, knock-in mice allow vaccination studies in a convenient, relatively low-cost mouse model. One can seed B cells from these mice at physiologically low numbers in otherwise WT mice and assess their responses in the face of competition from non bnAb clones. In these cell lines and mice, we have been carrying out studies to evaluate whether the imputed gl-bnAbs develop normally and to assess various vaccine candidates and immunization strategies. We have tested so-called germline-targeting immunogens engineered to bind better to gl-bnAbs than immunogens derived from WT Env. In the present proposal we focus on two aspects of vaccine design: the use of validated, rather than inferred, gl-bnAb precursors and an assessment of the ease of selectability of beneficial mutations using somatically-mutating human gl-bnAb cell lines and gl-bnAb mice. We hypothesize that by taking these approaches we will identify preferred gl-bnAbs that most effectively mature to become bnAbs and preferred immunogens or selection paths that should most effectively promote bnAb production when tested in primates. Because the problem of poor precursor affinity to antigen and uncertain paths to affinity maturation by somatic mutation are limitations to all immune responses, knowledge obtained here should be applicable to a variety of vaccine targets.